By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Almost every software company selling to freight now describes its product as AI. Some of it is useful. Some of it is the same software with a new label. As an AI transportation consultant, I think the buyer’s job is simple to describe and hard to do: find out what the tool does on your freight, with your data, when things go wrong.
These twelve questions will get you most of the way there. Ask them before you sign anything.
Questions About the Problem
1. What specific task does this replace or speed up?
A good answer names a task: reading delivery orders, booking terminal appointments, matching invoices to rate confirmations. A weak answer talks about transformation. If the vendor cannot name the task, you will not be able to measure the result.
2. How is that task done today, and what does it cost?
This one is for you, not the vendor. Count the hours and the errors before the demo. Without a baseline, every tool looks like an improvement.
Questions About Data
3. What data does the tool need from us, and in what condition?
AI built on incomplete or inconsistent data produces confident mistakes. Ask what fields are required, how clean they need to be, and who does the cleanup. I covered why this matters in AI in warehousing and 3PL operations.
4. Who owns our data once it is in your system?
Your rates, lanes and customer list are your business. Ask whether the vendor can use your data to train models that serve your competitors, whether you can export everything, and what happens to it if you leave.
5. Where is the data stored and who can see it?
Ask about access controls, encryption and whether any shipment or customer data is sent to third-party AI providers. Your customers may have contract terms that restrict this.
Questions About Accuracy
6. How accurate is it on freight like ours?
An accuracy figure from a demo data set tells you little. Ask for a trial on a sample of your own documents, loads or lanes, and check the output yourself.
7. What happens when it is wrong?
This is the most important question on the list. Does the tool flag low-confidence results for a person to review? Can you see why it made a decision? A system that is right most of the time and silent when it is wrong can cost more than the manual process it replaced.
8. Who is responsible for an error?
If the software misquotes a load, pays the wrong carrier, or misreads a hazmat document, your company is still the one accountable to the customer. Read the contract’s limitation of liability, and keep a person on anything with legal or safety consequences.
Questions About Fit
9. Does it connect to the systems we already use?
A tool that does not talk to your TMS, accounting system or ELD creates a new place to retype information. Ask for a list of live integrations, not planned ones. If you are still choosing a core system, start with what to look for in a TMS.
10. How long until it is working, and what do our people have to do?
Implementation is where most projects stall. Ask for a realistic timeline, the hours required from your team, and a reference customer of your size who went live recently.
Questions About Money
11. How is it priced, and what makes the bill go up?
Per user, per load, per document and per transaction pricing all behave differently as you grow. Ask what the cost would be at double your current volume, and whether there are setup or integration fees.
12. Can we start small and leave easily?
A vendor confident in its product will agree to a limited pilot with clear success measures and a short commitment. Be cautious about multi-year terms for a tool you have not seen work on your own operation.
How to Run a Fair Pilot
- Pick one task and one measure. Hours saved, errors reduced, or days to invoice.
- Use real work. Run it alongside your current process for a few weeks and compare.
- Check the exceptions. The easy cases will look fine. Look at the unusual ones.
- Ask the people doing the job. If dispatchers or clerks work around the tool, it is not saving time.
Warning Signs
- The vendor will not test on your data.
- Nobody can explain how the tool reaches a result.
- The answer to “what happens when it is wrong” is that it will not be.
- The contract is long and the pilot is not offered.
The Bottom Line
AI can remove real cost from a freight operation. It does that when it is aimed at a specific task, fed good data, and supervised by someone who knows the work. The twelve questions above are how you find out whether a product meets that standard before you pay for it.
For where the technology is already paying off, see AI in LTL and truckload shipping and AI in drayage and port trucking.
For more on freight and technology, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.